中国季度GDP核算数据质量研究  

Research on the Quality of China’s Quarterly GDP Accounting Data

作　　者：杨利雄[1] 张春丽[2] 李庆男 YANG Li-xiong;ZHANG Chun-li;LI Qing-nan(School of Management,Lanzhou University,Lanzhou 730000,China;School of Economics,Northwest Minzu University,Lanzhou 730030,China;Institute of Economics,National Sun Yat-sen University,Gaoxiong 80611,China)

机构地区：[1]兰州大学管理学院,甘肃兰州730000 [2]西北民族大学经济学院,甘肃兰州730030 [3]台湾中山大学经济研究所,中国台湾高雄80611

出　　处：《统计与信息论坛》2020年第2期3-9,共7页Journal of Statistics and Information

基　　金：国家自然科学基金青年项目“基于傅里叶近似的时变门槛模型:估计、检验及其应用”(71803072)

摘　　要：如何核算和发布季度GDP数据是一个主要难题。中国2017年改革了GDP核算方法,将季度GDP核算分为初步核算和最终核实两个步骤,取消了原先的初步核实,进一步研究取消初步核实的合理性。从理论角度,构造了一个新的等预测精度检验,新建立的检验具有更好的有限样本性质;从实证角度,首先基于季度GDP数据和第一、二、三产业的数据,排除了三大产业数据存在偏差相互抵消的可能,证实了初步核算数据以及初步核实数据均为真实数据的无偏有效估计,并基于新建立的检验比较了初步核算数据和初步核实数据的预测精度,发现初步核实数据预测精度更高,但两者预测精度差异并不显著,因此改革方案具有合理性。It is a world problem to calculate and publish the seasonal GDP data.The method of publishing China’s GDP data was reformed in 2017.Under the new accounting system,the preliminarily-verified data would not be published,and only the initial and final verification data are announced.The retionality of reform is assessed here and the results are put forward in twofolds.From a theoretical viewpoint,a new test for equal prediction accuracy is constructed,and the new test is showed to have better finite sample properties.From an empirical viewpoint,using the data of the headline GDP and the three sectors,the results show that there exist offsetting errors in the three sectors,and the preliminary and initial data are unbiased and efficient.Based on the proposed test,the accuracy of the preliminarily-verified and initial data are compared,which finds that the preliminarily-verified data are indeed more accurate,however,the accuracy difference is statistically non-significant.Hence,the reform of accounting GDP is reasonable.

关 键 词：GDP数据 初步核算数据 初步核实数据 无偏性 预测精度 

分 类 号：F222.33[经济管理—国民经济]